Health Insurance Fraud Detection Using Social Network Analytics

Abstract

Healthcare fraud detection is accomplished by mining social relationships and analyzing their patterns based on network data structures. Social networks are constructed which depict referral patterns (from health insurance claim information) and associations (from publicly available connection data) to analyze referral patterns and detect possible fraud, abuse and unnecessary overuse. The fraud and abuse management system supports the various aspects of fraud investigation and management, including prevention, investigation, detection and settlement. Using a unique combination of data mining capabilities and graphical reporting tools, the system can identify potentially fraudulent and abusive behavior before a claim is paid or, retrospectively, analyze providers' past behaviors to flag suspicious patterns.

1 . A computer implemented method for health insurance fraud detection comprising the steps of:
 constructing a social network database of physicians and health care providers from multiple, disparate and heterogeneous data sources;   extracting data pertaining to referrals between physicians and health care providers from a claim database; and   analyzing extracted data pertaining to referrals using data from the social network database to detect patterns that help determine scenarios for investigation.   
   
   
       2 . The computer implemented method for health insurance fraud detection recited in  claim 1 , wherein the step of constructing a social network of physicians and health care providers uses data mining techniques to find relationship data between physicians and health care providers. 
   
   
       3 . The computer implemented method for health insurance fraud detection recited in  claim 1 , wherein the step of constructing a social network of physicians and health care providers identifies physicians and health care providers as nodes with linkages in a social network, wherein the linkages are deterministic and probabilistic attributes. 
   
   
       4 . The computer implemented method for health insurance fraud detection recited in  claim 3 , further comprising the step of generating a graphical representation of the social network of physicians and health care providers. 
   
   
       5 . A system for health insurance fraud detection comprising:
 a plurality of disparate, heterogeneous data sources;   a social network analysis unit which receives input from said plurality of disparate, heterogeneous data sources and identifies relationship data between physicians and health care providers;   a social network optimization unit which receives input from said social network analysis unit and said plurality of disparate heterogeneous data sources and constructs a social network of physicians and health care providers;   data extracting means for extracting data pertaining to referrals between physicians and health care providers from a claim database; and   data analysis means for analyzing the extracted data pertaining to referrals using the social network to determine scenarios for investigation.   
   
   
       6 . The system for health insurance fraud detection recited in  claim 5  wherein said data sources include automated collection and user-generated data sources for social network construction. 
   
   
       7 . The system for health insurance fraud detection recited in  claim 5 , wherein the constructed a social network of physicians and health care providers identifies physicians and health care providers as nodes with linkages in a social network, wherein the linkages are deterministic and probabilistic attributes. 
   
   
       8 . The system for health insurance fraud detection recited in  claim 7 , further comprising means for generating a graphical representation of the social network of physicians and health care providers, which graphical representation also depicts referral patterns. 
   
   
       9 . A computer readable medium having code which implements a method for health insurance fraud detection, the method comprising the steps of:
 constructing a social network database of physicians and health care providers from multiple, disparate and heterogeneous data sources;   extracting data pertaining to referrals between physicians and health care providers from a claim database; and   analyzing extracted data pertaining to referrals using data from the social network database to detect patterns that help determine scenarios for investigation.   
   
   
       10 . The computer readable medium having code which implements a method for health insurance fraud detection recited in  claim 9 , wherein the step of constructing a social network of physicians and health care providers uses data mining techniques to find relationship data between physicians and health care providers. 
   
   
       11 . The computer readable medium having code which implements a method for health insurance fraud detection recited in  claim 9 , wherein the step of constructing a social network of physicians and health care providers identifies physicians and health care providers as nodes with linkages in a social network, wherein the linkages are deterministic and probabilistic attributes. 
   
   
       12 . The computer readable medium having code which implements a method for health insurance fraud detection recited in  claim 11 , further comprising the step of generating a graphical representation of the social network of physicians and health care providers.